Linux RDMA and InfiniBand development
 help / color / mirror / Atom feed
From: Leon Romanovsky <leon@kernel.org>
To: Boshi Yu <boshiyu@linux.alibaba.com>
Cc: jgg@ziepe.ca, linux-rdma@vger.kernel.org,
	chengyou@linux.alibaba.com, kaishen@linux.alibaba.com
Subject: Re: [PATCH for-next v3 0/3] RDMA/erdma: Add DMA-BUF memory registration
Date: Sun, 5 Jul 2026 13:08:41 +0300	[thread overview]
Message-ID: <20260705100841.GA15188@unreal> (raw)
In-Reply-To: <20260618045120.51210-1-boshiyu@linux.alibaba.com>

On Thu, Jun 18, 2026 at 12:51:02PM +0800, Boshi Yu wrote:
> Hi,
> 
> This patch series adds DMA-BUF memory registration support to the erdma
> driver:
> 
> - #1 renames get/put_mtt_entries to erdma_mem_init/uninit to better reflect
>      their purpose of initializing the struct erdma_mem.
> - #2 introduces struct erdma_mem_init_attr to pass parameters to
>      erdma_mem_init(), improving code maintainability and preparing for
>      DMA-BUF support.
> - #3 implements erdma_reg_user_mr_dmabuf() to enable DMA-BUF based user
>      memory registration using the revocable pinned dmabuf interface.
> 
> One known limitation: if the DEREG_MR command fails in the revoke
> callback, the driver can only log an error, since erdma does not
> support a function-level reset like irdma's request_reset() yet.

I hoped someone would comment on this. If revocation fails, the
hardware may continue accessing memory that the system considers
released.

From comment:
 * When a revocation occurs, the revoke callback will be called. The driver must
 * ensure that the region is no longer accessed when the callback returns. Any
 * subsequent access attempts should also probably cause an AE for MRs.

> We plan to introduce such a mechanism in future work.

That mechanism needs to be included in this series.

Thanks

> 
> Thanks,
> Boshi Yu
> 
> ---
> 
> v3:
>  - Patch#3: Switch to the revocable pinned dmabuf interface as
>    suggested by Jason Gunthorpe.
> 
> v2:
>  - Patch#2: Add validation for the return value of ib_umem_find_best_pgsz().
>   link: https://lore.kernel.org/all/20260518120637.16831-1-boshiyu@linux.alibaba.com/
> 
> v1:
>   link: https://lore.kernel.org/all/20260507053437.46211-1-boshiyu@linux.alibaba.com/
> 
> Boshi Yu (3):
>   RDMA/erdma: Rename get/put_mtt_entries to erdma_mem_init/uninit
>   RDMA/erdma: Introduce struct erdma_mem_init_attr
>   RDMA/erdma: Implement erdma_reg_user_mr_dmabuf
> 
>  drivers/infiniband/hw/erdma/erdma_main.c  |   1 +
>  drivers/infiniband/hw/erdma/erdma_verbs.c | 262 +++++++++++++++-------
>  drivers/infiniband/hw/erdma/erdma_verbs.h |  17 ++
>  3 files changed, 201 insertions(+), 79 deletions(-)
> 
> -- 
> 2.46.0
> 
> 

  parent reply	other threads:[~2026-07-05 10:08 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-06-18  4:51 [PATCH for-next v3 0/3] RDMA/erdma: Add DMA-BUF memory registration Boshi Yu
2026-06-18  4:51 ` [PATCH for-next v3 1/3] RDMA/erdma: Rename get/put_mtt_entries to erdma_mem_init/uninit Boshi Yu
2026-06-18  4:51 ` [PATCH for-next v3 2/3] RDMA/erdma: Introduce struct erdma_mem_init_attr Boshi Yu
2026-06-18  4:51 ` [PATCH for-next v3 3/3] RDMA/erdma: Implement erdma_reg_user_mr_dmabuf Boshi Yu
2026-06-18 14:27   ` Jacob Moroni
2026-07-05 10:08 ` Leon Romanovsky [this message]
2026-07-06  5:44   ` [PATCH for-next v3 0/3] RDMA/erdma: Add DMA-BUF memory registration Boshi Yu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260705100841.GA15188@unreal \
    --to=leon@kernel.org \
    --cc=boshiyu@linux.alibaba.com \
    --cc=chengyou@linux.alibaba.com \
    --cc=jgg@ziepe.ca \
    --cc=kaishen@linux.alibaba.com \
    --cc=linux-rdma@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox